application of total run-out

[Bill B.]

Total run-out interpretation.
I’ve been lurking on this forum for several months and I am impressed with the wealth of knowledge the members bring to the forum. My understanding of the standard has grown as a result. The standard I am referencing in the example is the ASME-Y14.5-2009. For background let me state that the part design is controlled by our customer, with one exception, that is the size of the spherical radius which is under our design control. As such the customer could not use profile to control the spherical radius and instead chose to use total run-out.
My understanding is that run-out, both circular and total, only control radial elements of the feature and not axial. However 9.4.2.1 states that total run-out may be used to control profile of a surface, which has me questioning whether my understanding is correct.
Looking at the example I have several questions:
1. What would a datum simulator for B,A look like? Is this a valid call out or should it be either A,B or just B ( there are no other FCF’s that use B,A so the intent would not be to create a simultaneous requirement)
2. Is this a legitimate use total run-out
3. If #2 is yes, Does total run-out control the 1.00 basic from datum A

I would like to thank everybody in advance for your comments and help.

 

[chez311]

Evan,

I agree on pretty much all points. There are special cases where uniform offset from the true profile is indistinguishable from translation.

We need to see what happens in a more general case (such as the curved surface in Fig. 12-10) to really know what the effect of a planar datum feature is. If total runout was specified on this feature, would the zone be allowed to translate relative to datum D? We don't know, because there is no example in Y14.5 with a feature like this (intentionally, I believe).

It occurred to me that the tolerance zone defined by circular runout when applied to a complex feature as in Y14.5-2018 fig 12-10 could also be affected by constraint of translation even though it only controls circular elements the tolerance zone is constructed from line elements normal to the true profile (and if that true profile is constrained in axial translation) - or what I assume is essentially the geometric/mathematic equivalent of the true profile though the terms utilized between Y14.5 and Y14.5.1 vary including toleranced surface and desired contour (the new Y14.5.1 draft comes closest with "true geometric shape"). I'm not sure the figures in Y14.5 shed much light on this, even though of course there are examples of a complex contour with circular runout. I may be off base though, what do you think?

Like I said above, I sought out the math standard hoping for some clarity but I was thrown off by the way the terms translated and scaled are used in Y14.5.1-1994 para 6.7 "The desired contour may be translated axially and/or radially, but may not be tilted or scaled with respect to the datum axis." It seems this remains in the draft, and I'm not really sure what to make of it if I'm being honest. Are these terms being used with a different definition than what I expect?

 

[3DDave]

Runout controls form variation, not nominal form or form location. Total runout controls variation in radius or axial location. Total runout cannot detect variations when both radius and axial location are changing as they two inputs cause measurements to vary in ways that cannot be directly attributed to either one unambiguously.

For the cone, is a detected variation due to an error in radius or an error in axial location?

With cylinders the measure is based on a constant radius and can detect radial errors. With perpendicular planar surfaces the measure is based on a constant axial location and can detect axial errors. In either of these two cases there is no requirement for any particular base value to make the comparison. One can set a radius of 1 inch or 10000 inches or an axial location of 1 inch or 10000 inches, as long as it is outside the material of the part, and record variations from those locations.

Trying to apply it to anything else means setting up a particular path to combine changes in radius, axial location, and the potential for orientation. And that means that each

For a radiused feature, that path varies from identical to the surface and to a degenerate path that has to located the exact center of the radius, to a path that is an inversion of the radius through the center of the radius; all while maintaining the measurement normal to the nominal surface.

The problem is that for any surface available a path can be found to minimize detected variations because there is nothing that defines the measurement path as basic, which seems to be a supported conclusion as these surfaces are usually defined with directly toleranced dimensions. That path will just be the locus of circular runout minimums at some user-chosen offset, so it is not different than a circular runout requirement.

 

[pmarc]

Bill's question was this: "The only thing I'm not clear on is the distance from the face (1.00 basic) controlled using total runout?".

So does the total runout on his sketch (assuming total runout is a legal control at all) control how deep the spherical feature can go into the material of the part, or not?

In other words, if the as-produced feature was shifted say .100 towards the material of the part, would/could this be caught during inspection of the total runout requirement?

 

[3DDave]

Since total runout is a subset of individual elements of circular runout and circular runout cannot detect the location of the surface, then the set that encompasses all potential circular runouts meeting a particular range cannot do so either.

 

[greenimi]

For the same reasons presented above even ISO allows only circular runout on conical or curved surfaces and not the total runout.
I took a quick glance in ISO1101:2017 and all shown cases for non-planar and non-cylindrical surfaces defined with runout are depicted with circular runout and not with total runout.
(unless of course if I am not missing something)

All total runout cases are shown on either cylindrical surfaces either a planar surfaces nominally perpendicular to the datum axis.

The one difference I've noticed for circular runout is that in ISO the angle should be TED, but ASME still allows it (the angle) to the ±.

If pmarc is still around, I will ask him to "slap me" if I said something incorrectly.

Thank you

 

[chez311]

Total runout cannot detect variations when both radius and axial location are changing as they two inputs cause measurements to vary in ways that cannot be directly attributed to either one unambiguously.

For the cone, is a detected variation due to an error in radius or an error in axial location?

Since total runout is a subset of individual elements of circular runout and circular runout cannot detect the location of the surface, then the set that encompasses all potential circular runouts meeting a particular range cannot do so either.

Even if we assume runout involves translation constraint of the true profile* in the axial direction, you are correct it may not be able to attribute/detect variation to one or the other unambiguously. I'm not sure how that comes into play though, plenty of controls cannot by themselves attribute variation unambiguously to one factor while still being affected by variations in said factor. Say for example parallelism applied to a flat surface where the surface is convex and takes up the entire tolerance zone - it would only be by evaluation of flatness that one could determine this variation is due to form error.

For a cone (or planar surface constructed normal to the datum axis) this is a special case, regardless of whether we assume translation constraint of the true profile in the axial direction, we can actually answer that question - the detected variation will never be due to an error in axial location as the tolerance zone can always be offset to accommodate it. I do not believe axial location error of a cone (or planar surface) will ever have an impact on total runout variation no matter how one interprets it.

With perpendicular planar surfaces the measure is based on a constant axial location and can detect axial errors.

If by "axial errors" you mean errors in form and orientation then I agree. If you mean axial location then I disagree - see my above, a planar feature would be one of these special cases.

An interesting point is that you mentioned the measurement is based on "constant axial location". Lets imagine a setup for measuring total runout of a planar feature that is normal to the datum axis. The indicator base is fixed at some axial location relative to the surface and swept across the feature surface in the radial direction. Theres no need to adjust the axial location as I noted above it has no impact on the measurement. Now if we take a complex shape such as the convex surface in pmarc's fig 12-10 from Y14.5-2018 if we assume no axial translation constraint of the true profile the indicator may have to be adjusted axially in order to get the part to fall within the tolerance zone. If we assume axial translation constraint of the true profile the indicator will not be adjusted axially (fixed in relation to some datum feature, but swept along a path defined by the true profile) and a feature could be rejected due to axial location error.

*I know that often these features will be directly toleranced, and a true profile requires basic dimensions - strictly speaking as defined in the standard. Substitute here if you wish any of the standard's alternate terms for this concept in these cases - toleranced surface, desired contour, and true geometric shape.

 

[chez311]

In other words, if the as-produced feature was shifted say .100 towards the material of the part, would/could this be caught during inspection of the total runout requirement?

If we assume axial translation constraint of the true profile*, and if the size of the feature is controlled to something sufficiently tighter than 0.100 then yes, a part shifted 0.100 will be rejected. Its worth noting that OP's case utilizes a MAX notation - if theres no minimum limit to size (probably not desired) then a feature shifted 0.100 towards the material would still be rejected (would require size variation greater than 2.50) but a feature shifted 0.100 away from the material could pass with a combination of location and size variation (size variation less than 2.50).

If we do not assume such axial translation constraint of the true profile then no, shift of the feature of any amount would not be detected as the indicator could be shifted axially along the axis to find a minimum value for total runout.

*See my above note (15 May 20 16:47)

 

[pmarc]

greenimi,
You are right that ISO 1101:2017 does not show an example of total runout tolerance applied to a surface other than regular cylinder or a planar surface perpendicular to the datum axis.

But it also does not specify how the feature controlled with a circular runout tolerance should be defined. It just shows how the circular runout tolerance zones relate to the toleranced feature depending on a scenario. For example, it tells what to do in order to specify that a circular runout tolerance applied to any surface of revolution defines tolerance zones at some fixed angle, rather than at the angle following the toleranced geometry. This practice does not exist in Y14.5.

 

[axym]

Hi All,

Interesting discussion.

I agree that the circular runout tolerance on the curved surface in Fig. 12-10 introduces some "axial" considerations. Inspecting this feature would involve fixed-angle indicator sweeps at a series of axial locations. The indicator needs to follow the basic contour, one cross section at a time, so that the fixed angle at each axial location is correct. So all of the fixed-angle sweeps need to have the proper axial location relative to each other. It would be possible to define runout tolerances so that the fixed-angle sweeps also need to have the proper axial location relative to perpendicular datum feature C, but I don't think that Y14.5 does this. There is no indication in the text or figures that runout tolerance zones are fixed in the axial direction, so that they would control the location of the feature in the axial direction. The only datum that is ever mentioned is the datum axis. When a perpendicular datum feature is present, its only described role is to define the orientation of the datum axis. So I don't think that there is any direct evidence to indicate that the axial location of the sphere would be controlled in the OP example.

The runout section of Y14.5-1966 includes figures with a part that looks very much like the one in Fig. 12-10 of 2018. Interestingly, the figure includes total runout tolerances on the conical surface and the curved surface. They show the indicator sweeping along the basic contour. At some point (possibly 1974 or 1982, I'm not sure), Y14.5 decided that total runout application on these feature types should not be mentioned anymore.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[chez311]

Evan,

If I'm being honest, as this discussion progressed the less I felt confident there was as much evidence that axial location might be constrained by runout tolerances. I still wanted to explore it as I don't think the standard is very clear on the topic and it highlights an important potential difference between total runout and dynamic profile.

In relation to your comments about circular runout - can I assume that the below bolded portions of your statement and that from Y14.5.1-1994 essentially describe the same thing?

The indicator needs to follow the basic contour, one cross section at a time, so that the fixed angle at each axial location is correct. So all of the fixed-angle sweeps need to have the proper axial location relative to each other.

A surface conforms to a circular runout tolerance to if all circular elements of the surface conform to the circular runout tolerance for the same mating surface.

I'm still befuddled by the verbiage at the end of para 6.7 but I guess the intent is clear enough for me now without that.

 

[axym]

chez311,

I agree that the standard is not clear on the concept of axial constraint by runout tolerances. I also strongly agree that dynamic profile introduces important differences - it would not just duplicate the same results as total runout if applied to the same feature types. Primarily, the vagueness would be eliminated - we wouldn't have to guess at things like the effect of axial datum features, patterns and grouping, and simultaneous requirements.

I'm still befuddled by the verbiage in para. 6.7 ;^). I believe that this paragraph is communicating the idea that runout tolerance zones are centered on the datum axis and oriented to it, but can translate along the axis. There is also the description of the FIM concept in mathematical terms, in which the tolerance zone originates at the "desired contour" but can offset/progress relative to it. This is a difficult idea to describe in words.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[chez311]

Evan,

Glad to see I'm not the only one that is put through a loop with the wording in 6.7!

I believe that this paragraph is communicating the idea that runout tolerance zones are centered on the datum axis and oriented to it, but can translate along the axis. There is also the description of the FIM concept in mathematical terms, in which the tolerance zone originates at the "desired contour" but can offset/progress relative to it.

I agree and understand that what the section is attempting to convey, but at the very least its a strange way to use the terms involved.

"The desired contour may be translated axially and/or radially"(Y14.5.1-1994) is a strange way to say "runout tolerance zones are centered on the datum axis". One would expect it to say instead "The desired contour may be translated axially but not radially" as translation radially would imply to me that it is NOT centered on the datum axis.

"[the desired contour] may not be tilted or scaled with respect to the datum axis"(Y14.5.1-1994) is a strange way to say "[the runout tolerance zone] can offset/progress relative to [the desired contour]". One would expect it to say instead "may be scaled but not tilted with respect to the datum axis" as NOT allowing it to scale as the Y14.5.1 text states suggests to me that it is NOT allowed to offset/progress. My first thought was that the desired contour represents the true profile (or equivalent) and is not scaled/offset with the tolerance zone, but after reading the following statements I'm not so sure - it seems the "desired contour" should actually be scaled/offset with the tolerance zone: "when a tolerance band is equally disposed about this contour and then revolved around the datum axis, a volumetric tolerance zone is generated" and "along the desired contour with the line segment kept normal to, and centered on, the desired contour at each point." I surmise this is likely why the refinement "the desired contour as defined by the true geometric shape" is added in the 20xx draft.

Perhaps I'm overthinking it, but the language to me seems to suggest the opposite of what is actually desired.

 

[axym]

chez311,

I agree that the awkward use of terminology and language in the runout section make it even more difficult to understand. There are familiar terms used in unexpected and counter-intuitive ways as you pointed out, along with unfamiliar terms such as "desired contour" and "actual mating surface". It's quite a journey to get from the description of a 2-dimensional desired contour that can translate radially to a 3-dimensional tolerance zone that can progress. With only one figure in the entire section, there is also a lot to try to keep straight in your mind as you read the text.

It took me years of reading and re-reading the runout section to piece together what it means. Which isn't really anything earth-shattering - it's just restating Y14.5's indicator-sweep definitions as tolerance zone definitions, using mathematical terms and notation.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[chez311]

It's quite a journey to get from the description of a 2-dimensional desired contour that can translate radially to a 3-dimensional tolerance zone that can progress.

Even then its a stretch as radial offsetting/scaling may be identical to radial translation for simple/special cases as you noted (planar/cylindrical/cone) its certainly not for complex profiles (concave/convex/etc..) even in 2D, hence my issues with the use of the term "radial translation" but I see what you mean.

I guess I'll try not to get too lost in the minutiae trying to tease these things out - I definitely understand the core concepts. Thanks for taking the time to work through it with me.

Consistent and well defined use of terminology would be helpful - ie: those you pointed out "desired contour" and "actual mating surface"/"mating surface" to which have been added "true geometric shape" and "design geometry". The introduction of the latter two (of which it is notable that true profile is a subset of design geometry) I think is a step in the right direction however the connection between some of these (and the reason for utilizing so many different terms) seems a bit fuzzy to me.